Specificity and sensitivity of ranked predictions

Copyright information:Taken from "Prediction of replication origins"Genome Biology 2004;5(4):R22-R22.Published online 4 Mar 2004PMCID:PMC395781.Copyright © 2004 Breier et al.; licensee BioMed Central Ltd. This is an Open Access article: verbatim copying and redistribution of this article are permitted in all media for any purpose, provided this notice is preserved along with the article's original URL. The training set was removed from consideration before generation of this figure. Prediction accuracy is depicted visually as a function of rank. Each prediction was plotted in rank order and coded green if it matched a member of the evaluation set of probable origins or black if it did not, and plotted in rank order from left to right. The high concentration of matches in the top predictions is visible as large blocks of green on the left. Specificity, defined as 100% minus the false positive rate, and sensitivity, 100% minus the false negative rate, are plotted for ranked groups of predictions in cumulative increments of 50 for the first 700 predictions and then for the total ranked list of 1,900 predictions. The ORC/MCM set was used for evaluation. Sensitivity gradually increases, and specificity decreases, as predictions of lower rank are included.